Hi. In the previous video, we took a look at the classic IBM PC junior computer and how it was a big flop back in the 80s and I'll link it in that down below and at the end of this video if you haven't seen it. and one of the things I briefly mentioned in the teardown of this is how the game port a joystick controller circuitry actually worked. When I was tearing down the piece of B we saw a five five eight timer in there and I explained that this was quite common back in the day, but I Thought we'd take a deeper look at that because it's rather interesting how this works.

I think it's rather clever. so let's take a look at the classic analog joystick. this is the original IBM joystick from back in the day IBM didn't actually I Don't think they actually made this because the original IBM PC actually didn't come with a joystick or a game controller port as it was called back in the day, but you could actually buy a plug-in card which actually was just the game controller PC card and that became the de facto industry standard joystick interface for the PC from then until the end of time until like USB joysticks took over, so they lasted a long time. this game controller joystick interface and it started out just as that single plug in peripheral car which could control two different joysticks through a D15 connector and a Wye splitter cable.

but of course that took up an entire slot in your PC and will dedicate in the hole slot away. And game controller kind of sucked. So eventually are these multi combo IO IO cards took over where you could. Actually, you know they would integrate a serial port, parallel port, and joystick controller ports, real time clock and all this sort of stuff into the one expansion card to try and save slots.

But the circuitry remained exactly the same for like a long time. And there's probably other brand personal computers back in the day that used a similar type of analog joystick interface. And that's exactly what this is. It's actually an analog joystick.

It It wasn't just buttons, you could actually get real feedback and real fine control. if you moved it just a little bit, then you're you know you're a spaceship and just move a little bit or whatnot. The PC could actually read the position of the joystick on here, and the standard IBM ones had two buttons like this. It was exactly the same on the Tandy 1000, which is a basically a clone of the IBM RPC Jr..

and was massively more popular than the PC Jr.. and I've done a video once again linked in at the end of this where I design a turbo board for the Tandy 1000. That's a real old video, but it's interesting, so stick around and check that one out. But exactly how did these analog joystick controllers work and what circuitry was required to actually read them? Well, let's do the first thing.

Tear this one apart. I Can just move this and you can see that these two rotating arms on here would actually just connect up to just a regular carbon pot in there, a potentiometer, a variable resistor, and it. They'd connect directly onto the shaft in there and simply turn that pot so the resistance value can see that they're only using the to wire one here, so they're not our Center tap in that. but it's fine.

you can do it either way, and that's how they do it. so they need to read the resistance value on this pot. This joysticks actually quite interesting in that it has a a free mode. You see how it actually springs back in both directions, where you can actually move it to the corner and then you could go free like that and then it would be free to sort of stay.

It didn't spring back in that direction, but it would still spring back in that direction and you could do that for both of those or either/or axes like that. So that was actually a really neat thing on the analog. PC Joystick A hundred K Like that, and I'm pretty sure that was just like an industry-standard value and you'll notice that they've got a little bit of a body in here. They've got, uh, some capsules that go into the shield of the cable.

Perhaps just some noise suppression there. I Don't remember my Tandy 1000 joystick ever having that. So how do we read the value from this pot? Well, that's where it gets a bit interesting. Now, of course you might be thinking that the obvious solution? You've got your potentiometer here.

Just hook it up to your 5 volt rail and your ground. and then you just, you'll get 0 to 5 volts depending on the location of the joystick there. And then you'd simply feed it into an analog to digital converter and you get your 8 bit or whatever value out. And Bob's your uncle, right? Well, you've got to remember that this was like the late 1970s, early 1980s, and IDC's These were actually quite expensive back then, and you'd need four of them.

You'd need at least a four channel one to read a dual joysticks in the PC two pots per thing. And of course, there was no such thing as microcontrollers back then. Essentially what we take for granted these days of our pick our AVR microcontrollers and they've got building a DC's up to 12 bits. And like all these peripherals, building basically microcontrollers didn't exist.

Back then. we had micro processors, which is what all these pcs ran on and if you wanted ad says you had to buy a separate ADC chip and they'll quite expensive because you've got to remember something like say, the first sort of reprogrammable microcontroller, the Pic 16 C 84 which used a squared prom. None this stuff less rubbish that didn't come out until 1993 and that didn't even have a built in ADC So you know you take these things for granted these days, but you haven't always had these micros with all these peripherals and a DC's built in, especially back then, so the clever designers went well. We've got a cut: cost doesn't get much cheaper than a Triple Five timer.

Let's take a look so you should be familiar with the classic Triple 5 timer and the various building block circuits that come along with that, and it's been popular for what? four decades or something and they still must produce this thing. Found a Triple 5 timer doing little one-shot and other timing type applications in a lot of the older vintage pcs from the 70s and 80s. in this case, the classic one-shot monostable circuit. And you know, familiar with this classic building block circuit.

We've got our resistor up here. you connect the discharge and the threshold pins. I won't go through how the triple five of our timer works. Triple Five timer t-shirt my hand-drawn t-shirt linked in down below.

By the way, if you want to check it out either, recommend you do. And then a cap go into ground and basically we have a negative going trigger pulse coming in. so when your trigger pulse goes low like that, it starts the timer which will produce a single pulse, the pulse width of which the time of the pulse, width is determined via the RC time constant here. And of course, if you're a place the resistor here with the pot used inside the joystick here.

Bingo! You've got yourself a proportional time pulse variable time pulse that will vary based on the position of the joystick. And because you've got a PC which has various timers and things built-in we can actually time in software how long that pulse takes. It's neat, so there's no need to convert an analog value into a digital value using an ADC. You can simply do it using a timer based approach.

and PCs are really good at doing timing, but of course, you need four of these triple five timers. Well, you didn't want to use for separate chips. Bugger that. But of course the Triple Five time is available in both the single one which is a triple five also the Five Five Six which is the dual triple five, and Tada the Five Five Eight, Quad Triple Five time.

I'm not very often used these days and this is what you saw in the teardown of the IBM PC Jr.. which has the original IBM game card or whoever made the design the IBM game card. It used a Five Five eight timer which is four triple five timers in the one chip and it was super cheap and there's basically not much more you had to add to it, so it had the four channels we could hook up the four joystick pots here. a four capacitors going to ground, they'd all be the same value.

might be typically 10 nano farad's or something like that. depends on the the speed of the PC and the timing and the hundred K used up here and stuff like that and Bingo! You would actually tie all the triggers together like this because they were independent timers in here. but if you tied them all together then then they all start at once as we'll see in a minute why. And then you would get the different pulse links out depending upon the position of each one of the joysticks.

The X&Y on joystick one and the X&Y Enjoy! Stick to beautiful and all the PC has to do is read the time period from there to there that are there that are there and that are there. Easy Peasy Lemon Squeezy. But of course the PC being the PC. It uses a data bus in the case of the IBM PC.

Of course, because it used the 8 O 88 processor, it had an 8-bit data bus instead of the 8 O 86 processor which had a full 16-bit data bus. We have to somehow get this hooked up to the data base so that we can use the timer's in the PC to actually time it. So how do we do that? Easy. We just hooked it up to a latch here, buffer / latch and then we would of course do the address decoding.

I won't go into architectures of PCs and stuff like that, but basically every every peripheral everything. every I/o port or memory port on the PC needed a particular address so it'll have an address decoder here typically using discrete and you know NAND gates and or whatnot stuff like that so that the and decode the address which is specific to the joystick. It did enable the chip and you could read back the values on all the on all four of these timer outputs here. but of course the joystick also had two buttons on so you could use the other four inputs here for the two buttons.

And of course you'd have an 8-bit data latch which then would go off to the PC bus. But how do you trigger it? Well, you use the address, same address decode but instead of reading back the data, you would actually write it. Some PC right pin would actually go to the trigger so you just write something to that address. It doesn't matter what data you wrote cuz the data was useless.

All you wanted to do is trigger these so you'd write to the particular joystick address and then you continually read back and until the end of the time period. But the problem with this approach of course is that it requires software timing loops and that ins a real problem to Pcs as they got faster and everything else. So there were very I believe there were various schemes eventually to compensate for that sort of thing, and ultimately all of this stuff would be built into the motherboards instead of add-on cards to be built in to the motherboard chipsets and things like that. But of course you still used the classic Five Five Eight on the output.

But you know all this stuff typically started to get integrated into the chipsets as time went on. So there you go. I Think that's a rather cheap and clever way to do it. All you needed was a five five a a couple of hours and C's hooked it up to your D 15 connector for your joystick and your buttons over here and Bob's your uncle.

You could just have a little address decode and read it back. Very simple and of course if you want to degrade a precision in reading the position of the pot, you had to do finer and finer reads on your port so that you could read the pulse-width there with greater precision. and we can actually see the value change here if we move the joystick. but of course you could trim it.

You had these little trimmers on the front that you can move back and forth to Center that and you can see the physically the joystick actually sorry, the pot actually moving back and forth as I tweak that trimmer on the top so you know you might want to trim it to the center or something like that. And if I move the joystick, you can see that it went all the way with LBJ up to 120 K The spots weren't very precise back in the day and then they go down to zero like that so you'd get the fool mostly the full range of the pot there. And if we actually have a look at the IBM PC Jr. motherboard, we can actually see the triple 5 timer up in there as we saw in our previous teardown.

There it is the classic National Semiconductor genuine. None of that ripoff rubbish Five, Five, Eight quarter, Triple Five timer. You'll notice some some of the time in caps there because of course you didn't want your PC shorting out and having no resistance at all in series with your cap. so they'd put some resistors in series for that just to give you and like, a restricted n stop type value on your joystick when your joystick shorted out and went to zero.

And of course, due to the proximity to the 508 there, that's almost certainly the buffer right. Used to go onto the PC bus. your absolute classic Sim 4 LS are 2, 4, 4 and they'd have probably some of your decoding stuff up here as well. And of course I'd love to show you the timing on this.

Unfortunately, if we have a look at the outputs here, of course then it actually does nothing. It's actually not triggering the output. It's not giving any pulses whatsoever on there because those timers wouldn't actually turn on our trigger unless the software was actually writing to that address. So you're running a game or something that was continually polling the joystick.

So in this case, this PC Jr. doesn't work. Oh, now the Tandy 1000 is interesting in that it actually uses a slightly different Hardware implementation. It doesn't use the triple 5 timer, so let's actually take a look at how it actually does it.

Here's the block diagram here. It's essentially doing exactly the same thing as what we were doing with the triple 5 timer. It's producing a proportional pulse with timer output based on a trigger signal. But it's doing it using an integral here.

and that's what that little fancy symbol is integral of IDT there. What that means is that it's just producing a ramp which starts at 0 when you trigger it and that voltage ramps up. and if we have a look at the schematic here, you can see that that little ramp generator there with the J FET There it just generates a ramp voltage which then goes into our comparators there which then the joists. The voltage from the joystick just goes into these four comparators and then of course once the ramp voltage gets up and reaches each individual joystick position, the comparators flip over and produce a zero instead of a vice versa.

and then the PC can actually read that as a time period. Exactly the same as a triple five timer, except the hardware implementations. A little bit different, but it works exactly the same and you can see the seven four, LS, two four four buffer on there and the right and the address decoder and the right signal. and the read signals work as well.

It works exactly the same as the original IBM PC game adapter. In fact, the software as long as you kept the read and write address the same software wouldn't even know the difference. Wouldn't know whether using the integrator there like the ramp generator and the comparator or whether or not you using the triple five timer still going to work exactly the same software wouldn't notice the difference I accept if there's some variations in there in the actual timing itself. but apart from that not you get the same thing and I'd love to show you the Tandy 1000 working as well.

This is the IBM PC Jr. 1000 down in my storage bunker, but I can't find my five and a quarter inch boot floppies. so I just can't get the thing going. and without that software, there's nothing to trigger the polling or whatnot without even being able to boot up DOS to then you know, be able to write to the joystick address to trigger the damn thing so that we can see the time and stuff like that.

But I hope they gave you a good idea of how the IBM PC joystick game adapter actually works. It's not the greatest solution because it is, you know, software timing-dependent You've got to be able to accurately measure the time periods and do it fast enough to respond to all the joystick controls and of course Pcs. Back then they had to share all the same bus and they were all doing. You know they had to do everything at once, but you know it, it actually worked back in the day and that was a fairly cheap and simple solution to bring the component cost down instead of using an ADC or something like that.

So anyway, I hope you enjoyed that and found it useful. If you did, please give it a big thumbs up. As always, discuss down below in the YouTube comments or on the EEV blog forum and there's a couple of videos here at the end you should check out and faced all my patrons and supporters who are keep this channel going. You can probably find a link here at the end somewhere too if you want to join anyway.

catch you next time.